Phototransduction in rod and cone photoreceptors utilises a G‑protein cascade that is well understood at a cellular and molecular level.  Rhodopsin (or its cone equivalent) is a typical G‑protein-coupled receptor and, when activated by light, enzymatically activates the G‑protein, transducin.  Transducin in turn activates a cyclic-nucleotide phosphodiesterase (PDE6), increasing the hydrolysis of cGMP, and thereby causing closure of cyclic nucleotide-gated channels in the cell’s plasma membrane.  The activation phase of the light response in rods has been modelled accurately at a molecular level.  An additional half-dozen proteins are involved in shut-off of the light response and light adaptation. Remarkably, the homologous cascades in rods and cones use different isoforms of most of the main players.  This duality can be traced back to the two rounds of whole-genome duplication (WGD) that occurred roughly 500 million years ago in a chordate ancestor of all jawed vertebrates, or in some cases to more ancient duplications (Lamb, 2020).  For transducin a, for the PDE6 catalytic and regulatory units, and for the CNGC a and b subunits, this rod versus cone distinction arose at the first round (1R) of WGD.  The HGNC names of the respective rod / cone isoforms of these genes are: GNAT1 / GNAT2; PDE6A+PDE6B / PDE6C; PDE6G / PDE6H; CNGA1 / CNGA3 and CNGB1 / CNGB3.  For the visual pigments, three cone-type opsins with different spectral sensitivities had already evolved prior to 1R, along with pinopsin, which appears to have been the ancestral scotopic visual opsin; rhodopsin did not appear as a distinct isoform until 1R.  Likewise, most of the other proteins involved in response recovery and light adaptation exhibit distinct rod/cone isoforms, and they too arose through a combination of ancient (pre-1R) individual gene duplications and the first round of WGD. The existence of these multiple isoforms underlies some marked differences in rod/cone manifestations in a variety of monogenic retinal disorders. Recently, it has become apparent that the dimeric nature of the rod PDE6a/b plays an important role in enabling the rod to operate at very low intensities.  Activation of the two catalytic subunits exhibits significant cooperativity, in that a single bound transducin elicits only a small fraction of the activity elicited when two transducins bind.  This cooperativity provides immunity against spontaneous thermal activation of transducin, because appreciable PDE6 activity occurs only upon a coordinated burst of transducin activation.  Such a noise-reduction mechanism appears crucial in enabling the rod to respond reliably to a single photon.  However, this benefit comes at the cost of a short delay (of ~5 ms); this is probably immaterial to rods, but would be a severe disadvantage if it were to occur in cones which mediate responses at high temporal frequencies.  Analysis of the dimeric PDE6 activation scheme provides a quantitative explanation for rod recovery from intense flashes, and for the occurrence of an increase in the rod’s ‘dominant time constant’ at high intensities.